An Effective Approach for Mining Weighted Sequential Patterns

Abstract

Sequential pattern mining is one of the most studied data mining problem and has wide range of application domains including weather prediction, network intrusion detection, web access analysis, customer purchase analysis, etc. The weighted sequential pattern mining is an approach to find only interesting sequential patterns by assigning weights to data elements present in the sequences. The time-interval weighted sequential pattern mining is another approach in which different weights are assigned to the time-interval values between the successive transactions. From customer purchase pattern analysis point of view, both item’s importance as well as time-interval gap values is useful and more interesting patterns can be discovered by considering them while assigning weights to the sequences. This paper aims to propose a novel approach for finding weighted sequential patterns from customer retail database which incorporates both the item’s importance and time-interval gap information so that the discovered sequential patterns will be more meaningful and effective for the end-user. The results infer a lot of computation cost can be saved by focusing on few interesting patterns.